EYE BLINK DETECTION USING ALGORITHM BASED ON dlib AND openCV LIBRARY FOR GAME PLAYERS IN COMPETITIVE ENVIRONMENTS

Rationale: Research on the correlation between brain activity and blinking is very important in finding a way to improve non-invasive diagnostic methods for brain diseases. Parkinson’s disease has a 29% misdiagnose rate, which becomes even higher during the early stages of the disease. As a result, a nonintrusive way of measuring brain activity is needed. Among several different biometrics, the eye blinking rate was recognized to have correlation with complex activities in the brain.

Procedures: This experiment measures the blinking rate of adolescents, who are maturing their prefrontal cortex using algorithm based on dlib and openCV library. Eye blinking was measured during the playing of a game, where various decisions must be made in a short amount of time. Through this measurement, the number of blinks in non-competitive and competitive situations was compared, and determined which situation results in faster blinking through the Eye Aspect Ratio (EAR). In addition, it could be examined on which objective affects brain activity the most, by measuring the number of blinks and EAR in different genres of games.

Results: The optimal threshold value for the blinker program was 0.18. The range of the slope of the real-time recording of the blinking counter is about 4 times larger in the competitive mode. These results support eye blink count and EAR varied in all games in the competitive mode. This suggests that each situation in the competitive games was unique and required different amounts of brain activity. It was inferred that the brain worked actively in these stressful situations.